Overfitting

How it is used

A classic warning sign is training loss going down while validation loss goes up. The verb is overfit: a model overfits, or overfits to the training set. Engineers fight it with more data, regularization, dropout and early stopping.

Example sentences

1. Validation loss started rising after epoch 8, a clear sign of overfitting.

2. With only 300 examples, overfitting is almost guaranteed.

Common collocations

• reduce overfitting
• prevent overfitting
• signs of overfitting
• overfit to the training data

Related terms

• underfittingA problem in which a model is too simple or trained too little to capture the patterns in the data, so it performs poorly even on the training data.
• regularizationAny technique that stops a model from becoming too complex during training, in order to reduce overfitting and help it work better on new data.
• validation setA part of the data kept out of training and used during development to compare models, tune settings and decide when to stop training.
• training dataThe examples a model learns from during training. Their quality, size and variety largely decide how well the model works on new data.

This word in other fields

The same English word can have a different meaning in another field:

• overfitting · Software Engineering
• overfitting · Data Science & Analytics

More terms: Model Training

• train
• loss function
• epoch
• learning rate
• hyperparameter
• batch size
• gradient descent
• checkpoint
• converge